Our previous studies have shown that relatively early-stage Alzheimer's disease (AD) patients have already endured a significant deterioration in the structure of their semantic knowledge. Given that AD is a progressive disease, the patients' semantic network should manifest a further deterioration by the later stages of their disease. To evaluate the nature and degree of this deterioration, the present study will use multidimensional scaling techniques (MDS) to compare the semantic networks of mildly (Mid-AD), moderately (Mod-AD) and severely (Sev-AD) demented AD patients. It is anticipated that the semantic networks of Mid-AD, Mod-AD and Sev-AD will be different in terms of the primary dimension of organization, the clustering of concepts and in the variance or heterogeneity of performances within each group. Understanding the semantic networks of AD patients in different stages of the disease is both theoretically and clinically significant. Given that AD results in severe pathological changes in the cortex, examining the semantic networks of AD patients will help in understanding the different neuronal pathways by which different components of semantic information are stored and distributed. On a clinical level, a thorough knowledge of the progressive changes in AD patient's semantic space could have importance for the early detection and accurate staging of the disease.